Credit as the Insurance Nobody Sells: Why Borrowing Absorbs Shocks Badly

A transmission fails. Three households face the same $2,800 repair. The first pays a deductible because the risk was insured. The second writes a check because the money was saved. The third borrows, and pays $2,800 plus interest over eighteen months while their available credit sits consumed. Same event, three mechanisms, and which one a household uses is determined by resources rather than by which works best. The third is the worst of the three by construction — insurance transfers risk, savings pre-funds it, and credit only defers it. Much of what we call consumer borrowing is households doing insurance's job with an instrument that isn't insurance, and the substitution fails hardest at exactly the moment insurance would perform.

Three mechanisms, ranked

Households face shocks — income interruptions and unexpected expenses. There are exactly three ways to absorb one:

	Insurance	Savings	Credit
What it does to the risk	Transfers it to a pool	Pre-funds it	Defers it
Household bears	Premium plus deductible	The full cost, once	The full cost plus interest
Requires qualifying	At purchase	No	At the moment of need
Available in a downturn	Yes — that's the design	Yes	Contracts
Consumes future capacity	No	Yes, until rebuilt	Yes, for the loan's life
Cost if the shock never happens	The premiums	Forgone return	Nothing

Read the bottom row before concluding credit is simply worst — it has one genuine advantage, and we return to it. But on every row concerning what happens when the shock arrives, credit is dominated.

The third row is the one that does the most damage and gets the least attention. Insurance and savings require no approval at the moment of need. Credit does. A household relying on borrowing has an absorption mechanism contingent on a third party's assessment of them, made under the conditions the shock created — which is the timing trap our illiquidity analysis identifies in home equity, generalized to all credit.

Our thesis in one sentence: a substantial share of consumer borrowing is not life-cycle consumption smoothing but shock absorption performed by an instrument poorly suited to it, used because the better instruments are unavailable to the household.

Insurance and savings are available when you need them. Credit requires someone to approve you at the moment the shock has just made you look worse.

What the same shock costs

Put numbers on the $2,800 repair.

	Insured	Saved	Borrowed at 26%	Borrowed at 90%+
Immediate outlay	$500 deductible	$2,800	$0	$0
Interest paid	—	—	~$690 over 18 months	~$2,400+
Annual premium	Say $340	—	—	—
Total cost of the event	~$840	$2,800	~$3,490	~$5,200
Capacity consumed	None	Buffer, rebuildable	18 months of credit line	Months of income

The spread is roughly six to one between the best and worst mechanism for an identical event. And the household using the worst one is the one least able to absorb it.

Three things this table understates.

The capacity cost is real and unpriced. A household that borrows $2,800 has consumed capacity that is now unavailable for the next shock — and shocks are not one-per-lifetime. The second shock arrives with the first still on the balance sheet, which is the cascade our revolver analysis describes: the mechanism used to absorb one event is what makes the next unabsorbable.

The rate isn't independent of the household. Rates rise as credit standing falls, so the households most likely to face uninsured shocks face the highest cost of absorbing them — the compounding our poverty premium analysis documents.

The insured column requires having bought insurance in advance, which requires the foresight and the cash flow to pay premiums against an event that may not happen. That's a real requirement and it's the reason insurance isn't simply the answer.

The failure at the worst moment

The structural defect that distinguishes credit from insurance most sharply, and it's the analytical core of this report.

Insurance is designed to pay out when claims arrive. Credit contracts when claims would.

Consider what happens in a widespread shock — a regional employer contracting, an energy cost spike, a downturn:

• Many households need to absorb a shock simultaneously.
• Lenders observe rising delinquency and deteriorating conditions.
• Limits are reduced, criteria tighten, approvals fall.
• The absorption mechanism withdraws precisely as demand for it peaks.

This is the same common-factor structure our correlation analysis describes from the lender's side, viewed from the household's. The lender's correlated losses and the household's unavailable credit are the same phenomenon: consumer credit risk doesn't diversify because households share exposures, and the credit that would absorb those exposures is priced and rationed on the same information.

An insurance pool works because it's capitalized in advance against correlated claims. A credit line has no such capitalization — it's a discretionary commitment that shrinks when the issuer's own risk assessment worsens. The two instruments respond to correlated stress in opposite directions, which means substituting one for the other converts an insurable risk into an uninsurable one.

The lender's parallel failure is worth naming: this is also when the forbearance option is most valuable and least exercised, because the same conditions that make borrowers need relief make institutions least willing to grant it. The system tightens at both ends simultaneously, and the household experiencing it as bad luck is experiencing a design property.

Why these risks aren't insured

The obvious response — insure them — deserves a serious answer, because the gap is a genuine market limitation rather than simply a failure of provision.

The shocks households actually face:

Shock	Insurable?	Obstacle
Major medical event	Yes, largely	Coverage gaps and deductibles
Home structural damage	Yes	Cost and availability, per our premium analysis
Vehicle collision	Yes	Deductible only
Hours reduced	Barely	Frequent, small, hard to verify, influenced by the insured
Vehicle mechanical failure	Poorly	Maintenance versus failure is contestable
Appliance failure	Poorly	Administration cost exceeds the payout
Gap between jobs	Partially	Unemployment insurance excludes much of the workforce
Irregular income	No	The variation is the job

Read the bottom half. The shocks that drive most emergency borrowing are the ones that are hardest to insure, and for reasons that are structural rather than incidental:

• High frequency, low severity. Insurance administration has a fixed cost per claim, and a $600 repair can cost more to adjudicate than to pay — the same fixed-cost-against-size problem our selection analysis identifies in small-dollar lending. Small risks are hard to insure for exactly the reason small loans are hard to make.
• Verification difficulty. Whether hours were cut or a shift was declined is hard to establish independently.
• Moral hazard. Coverage against outcomes the insured influences invites the outcome.
• Adverse selection. Households expecting volatility buy volatility coverage.
• Correlation. Employment shocks hit pools together, which is expensive to capitalize against.

Which yields an honest and uncomfortable conclusion: credit filled this gap because insurance genuinely couldn't, not because nobody thought of it. The market limitation is real, and any proposal to insure these risks has to solve the five problems above rather than assert that the coverage should exist.

What this reframes

Several things the library has documented look different once you treat emergency borrowing as attempted insurance.

Small-dollar credit demand isn't primarily consumption. The products in our selection analysis are frequently described as funding overconsumption. Most of the borrowing is shock absorption — the demand exists because an insurance market doesn't, and the moral framing of the borrower misses the structure entirely.

Emergency savings is under-valued by exactly this analysis. Our savings analysis treats a buffer as prudence. It's better understood as self-insurance — the household holding the reserve is capitalizing its own pool of one, which is inefficient relative to a real pool and vastly superior to borrowing. This is why a modest buffer produces effects out of proportion to its size, and it's why the finding that small dedicated savings dramatically reduce retirement cash-outs in our leakage analysis is the most important number in that report.

Earned wage access is an income-timing product, not a loan in economic function — which is why the classification fight in our advance analysis is genuinely hard. It addresses a timing mismatch, which is closer to what the household actually needs than credit is.

The illiquid household is uninsured, not imprudent. A household with substantial net worth and no buffer has assets that can't absorb shocks. They're wealthy and uninsured simultaneously, which is the specific condition that produces retirement withdrawals and revolving balances among people who look comfortable.

Forbearance is a form of insurance the lender can provide. Deferring payments during a temporary shock is risk-sharing, and the arithmetic in our forbearance analysis says it's frequently profitable for the lender too. The instrument closest to insurance in the existing system is a workout, and it's systematically under-supplied.

The honest case for credit

Credit has one genuine advantage over both alternatives, and it explains the persistence better than any account based on exploitation.

You don't have to specify the peril in advance. Insurance covers named risks; a policy against transmission failure doesn't pay for a medical bill. Savings is general but must be accumulated beforehand. Credit is general-purpose and requires nothing in advance — a household that never anticipated the specific event can still access it.

That flexibility is worth a great deal, and it's why credit isn't simply a failure state:

• No premium if the shock never comes. Insurance costs money whether or not you claim; credit costs nothing until used. For a household with severe cash flow constraints, avoiding a certain premium in favour of a contingent cost is a defensible trade.
• No accumulation period. Savings requires time the household may not have had.
• It covers uninsurable risks. Given the table above, credit is the only mechanism available for most of what households actually face.
• Speed. Immediate where insurance requires a claim process.

So the accurate statement isn't that credit is a bad shock absorber. It's that credit is a poor substitute for insurance and the only general-purpose absorber available — which makes its cost the price of a missing market rather than evidence of anything about the borrower.

What would dominate it

If the diagnosis is that households are using a deferral instrument for a transfer problem, the remedies follow from the diagnosis rather than from sentiment:

Automatic liquid savings. The highest-return intervention available, and the reason is the fixed-cost logic above. A buffer is self-insurance that requires no pool, no verification, no adjudication, and no qualification. Attaching it to payroll — where illiquid saving already happens by default — meets the household where the behaviour already is, which is why default-based accumulation outperforms exhortation by margins our disclosure analysis would predict.

Income smoothing rather than income lending. A product that moves income within a pay cycle addresses a timing mismatch without creating an obligation. That's structurally different from credit and better matched to the problem.

Employer-adjacent hardship provision. Employers can verify employment shocks cheaply — the verification problem that defeats insurers is trivial for the employer. This is the one place where the adjudication cost is near zero, which makes it the most plausible site for something insurance-like at small scale.

Better-designed workouts, which are risk-sharing between lender and borrower and are under-supplied for the measurement reasons our forbearance analysis identifies.

Cheaper credit where credit is unavoidable. Given that most of these risks are genuinely uninsurable, some of this demand will be met by borrowing regardless — which makes the fixed-cost interventions in our selection analysis directly relevant. Reducing the cost of the imperfect instrument is a legitimate goal alongside building better ones.

The strongest objections

"Much emergency borrowing isn't for emergencies." Fair. Households borrow for things that aren't shocks, and stated purpose is unreliable. Our response is that the distinction matters less than it appears: whether an expense is a genuine emergency or a poorly timed choice, the absorption problem is the same, and the mechanism used is still determined by resources. But we'd concede that the share of borrowing genuinely attributable to shocks is contested and we haven't quantified it.

"Insurance has its own failures." Strongly true and worth stating. Claims are denied, coverage has gaps, premiums rise, and the property insurance market our premium analysis describes is currently failing in availability as well as price. The comparison is between imperfect mechanisms, not against an ideal. Insurance is better than credit at absorbing shocks it actually covers, which is a narrower claim than insurance being good.

"This romanticizes savings for households that can't save." The sharpest objection. Telling a household with no margin that they should self-insure is telling them to do something the constraint prevents — which is why we'd emphasize automatic and default-based accumulation over advice, and why the honest conclusion includes making credit cheaper rather than only recommending its alternatives. A framework that concludes "these households should have saved" has restated the problem as a solution.

Testable implications

1. Emergency borrowing should fall where insurance coverage is broader, holding income constant — the substitution claim, and it's testable across coverage regimes.
2. Small liquid buffers should reduce high-cost borrowing by more than their size implies, because they prevent the capacity-consumption cascade rather than merely funding one event.
3. Credit availability should contract most for the households most exposed to common shocks — the correlation failure, and it predicts that access falls fastest where it's needed most.
4. Employer-provided hardship access should be cheaper than market credit at equivalent risk, because the verification cost is near zero.
5. Households with insurance should show lower credit utilization volatility, not just lower levels.
6. Borrowing for shocks should show different repayment behaviour than borrowing for planned purchases at the same amount and borrower quality — which, if true, is a usable underwriting signal and one that stated purpose would partly capture.

The sixth is the one with commercial as well as analytical value. If shock-driven borrowing performs differently from planned borrowing, then purpose carries information that most underwriting discards — and it's testable on any lender's existing book.

The conclusion we'd hold: consumer credit is doing a job it wasn't built for, because the instrument built for that job doesn't reach most households and in several cases can't. That reframes the policy question from how to restrict borrowing to how to reduce the need for it — and the answer isn't primarily about credit at all.

Frequently asked questions

---

This report presents an analytical framework and the authors' interpretation; it is not financial or insurance advice. Worked figures are stylized illustrations using assumed rates, premiums, and terms. The share of consumer borrowing attributable to shocks rather than planned expenditure is contested and is not quantified here.